Integrating remote sensing and GIS for environmental justice research - Urban Remote Sensing: Monitoring, Synthesis and Modeling in the Urban Environment - page 223

the region, i.e., even among the highly educated in the region,
minorities are still less likely to live nearby abundantly vegetated
areas. The role of population density suggests that this negative
relationship between minorities and vegetation is explained in
large measure by the degree of ''urbanness.'' One narrative inter-
pretation of these results would be that minorities tend to live in
more densely populated, urban areas; and because urban areas
tend to have lower vegetation concentration, minorities tend to
live in less green areas than non-minorities. This interpretation
suggests that racial inequity in vegetation character is a func-
tion of historical processes related to residential mobility. Racial
segregation in residential housing, as well as historical processes
of suburbanization and ''white flight,'' have produced the cur-
rent racialized spatial residential patterns that can be currently
observed (Pulido, 2000).

SMA has also been applied to multi-spectral IKONOS imagery
to yield distinct categories of vegetation types, such as trees and
grasses (Nichol and Wong, 2007).
Increases in spatial and spectral resolution are not only useful
for measuring vegetation character as an indicator of an environ-
mental amenity in environmental justice research, but also can
serve as useful ancillary data sets for dasymetricmapping of popu-
lation. For example, in the case study above, population data were
attached to Census tracts and the analysis makes the assumption
that population is distributed homogeneously throughout each
tract, whereas in reality (nighttime) population is concentrated
in residential parts of tracts. High spatial resolution imagery from
IKONOS has been used in dasymetric mapping to disaggregate
census population data and derive population distribution data
sets at a much finer resolution than the original census data,
andanevenfinerresolutionthanwouldbepossibleutilizing
Landsat or other more commonly used remotely sensed ancillary
data for dasymetric mapping (McKenzie, 2008). Because more
accurate population data can improve the estimation of spatial
relationships between population and environmental hazards
and/or amenities, these advances in the spatial and spectral res-
olution of remotely sensed data can contribute to advances in
environmental justice.
It is worth noting that remotely sensed imagery has also
been used not only for dasymetric mapping of population but
also for population estimation in the absence of a priori census
data. For example, Mubareka et al . (2008) use Shuttle Radar
Topography Mission (SRTM) and Landsat Thematic Mapper
data to estimate settlement location and population density in
northern Iraq using a small sample of settlements to calibrate
the population estimation, whereas others have used high res-
olution satellite imagery to simply count residential dwellings
and estimate population (e.g., Bjorgo, 2000). Such studies may
be particularly useful in estimating population in Third World
nations where census information may be very spatially coarse,
of unknown accuracy, not current, or simply nonexistent. One
notes, however, that these techniques may be difficult to extend
to enumerating not only population, but also indicators of race,
class, and other socioeconomic characteristics that form the basis
of environmental justice research.
Remotely sensed data may also be used to capture other
environmental amenities, besides vegetation, or hazards that
may be useful in environmental justice research. As noted earlier,
the majority of environmental justice studies have focused on
technological hazards from industry, for which data are typically
derived from environmental databases of toxic emissions and/or
environmental monitoring networks. As with the use of thermal
imagery for studying the environmental justice aspects of the
urban heat island effect, the use of remotely sensed imagerymay
be used to substantially expand the scope of environmental justice
research to a variety of types of environmental hazards that have
not been substantially addressed in the environmental justice
literature. Remotely sensed data can provide valuable data on
environmental risk due to natural hazards, including volcanic
eruptions, earthquakes, landslides, and flooding. High resolution
elevation data in particular, as derived from light detection and
ranging (LIDAR) and interferometric synthetic aperture radar
(InSAR), have proven particularly useful in estimating areas at
risk for a variety of types of hazards (Tralli et al ., 2005).
Many remote sensing technologies have now been in oper-
ation for years, and some for decades. Thus, remotely sensed
data on vegetation and other environmental characteristics can

Conclusion

Several prospects and challenges can be identified for the contin-
uing integration of remote sensing and GIS for environmental
justice research. One issue concerns how new remotely sensed
data products can aid environmental justice research, as with
recent advances in the spatial resolution of remotely sensed
image products. In the case study above, for example, vegetation
concentration was captured at a resolution of 30 m, potentially
masking substantial spatial variation in vegetation character.
Urban environments typically have high spatial heterogeneity in
land cover, which can change abruptly over short distances. For
instance, parkland (with high vegetation concentration) can lie
directly adjacent to industrial land with effectively no vegetation
present. Or, similarly, consider a main commercial street corri-
dor running through a suburban neighborhood. The suburban
area, with a single-family home style of residential development,
may have relatively strong vegetation concentration due to the
presence of lawn grasses and moderate tree cover. A main com-
mercial street running through such a neighborhood, however,
is likely to have far less vegetation and greater impervious surface
cover. In such cases as these, remotely sensed imagery at a reso-
lution of 30 meters is unlikely to capture the detail of such spatial
variation in vegetation, as the resolution is simply too coarse
to accurately capture abrupt changes in vegetation character (as
with the boundary between parkland and industrial land) and
narrow or other oddly shaped features in the landscape (such
as commercial streets that bisect residential neighborhoods, or
waterways).
Commercially available multi-spectral image products with
less than five meter resolution, such as IKONOS (Space Imaging,
Inc.) and Quickbird (DigitalGlobe, Inc.) imagery, allow for much
finer resolution and precise spatial estimates of urban vegetation
character (Nichol and Lee, 2005). Advances in spectral resolution,
on the other hand, allow for the recognition of not only general
measures of vegetation concentration, as with NDVI, but the
recognition of specific types of vegetation and other land covers.
Hyperspectral sensors carried on board airborne platforms, such
as Airborne Visible Infrared Imaging Spectrometer (AVIRIS),
which has 224 spectral bands, can provide imagery from which
specific types of vegetation may be extracted. SMA has been
applied to AVIRIS imagery to yield various types of vegetation,
including woody vegetation and grasses (Wessman et al ., 1997),
as well as to distinguish among other types of urban land covers.